Since the development of functional magnetic resonance imaging in the 1990s, reliance on neuroimaging has exploded as researchers explore how fMRI data from the resting brain and the anatomical structure of the brain itself can be used to predict individual characteristics , such as depression, cognitive impairment. and brain disorders.
Brain imaging has the potential to reveal the neural underpinnings of many traits, from disorders such as depression and chronic widespread pain to why one person has a better memory than another and why some people’s memories are resilient as they age. But how reliable brain imaging is for feature detection has been widely debated.
Previous research in brain-related studies (called ‘BWAS’) has shown that links between brain function and structure and traits are so weak that thousands of participants are needed to detect replicable results. Research of this scale requires millions of dollars of investment in each study, limiting what traits and brain disorders can be studied.
However, according to a new comment posted on Nature, stronger associations between brain measures and traits can be obtained when using state-of-the-art pattern recognition (or “machine learning”) algorithms, which can gather high-power results from modest sample sizes.
In their paper, researchers from Dartmouth and University Medicine Essen provide a response to a previous analysis of brain-wide correlation studies led by Scott Marek at Washington University School of Medicine in St. Louis, Brenden Tervo-Clemmens at St. Louis General Hospital Massachusetts/Harvard Medical School and colleagues. The previous study found very weak associations across a range of traits in several large brain imaging studies, concluding that thousands of participants would be needed to detect these associations.
The new paper explains that the very weak results found in the previous work do not apply to all brain images and all features, but are limited to specific cases. He describes how fMRI data from hundreds of participants, as opposed to thousands, can be better harnessed to yield important diagnostic information about individuals.
One key to stronger correlations between brain images and traits such as memory and intelligence is the use of state-of-the-art pattern recognition algorithms. “Since there is virtually no mental function that is performed entirely by one brain region, we recommend using pattern recognition to develop models of how multiple brain regions contribute to trait prediction, rather than controlling which regions brain individually,” says senior author Tor Wager. Diana L. Taylor Distinguished Professor of Psychological and Brain Sciences and director of the Center for Brain Imaging at Dartmouth.
“If we apply models of multiple brain regions that work together rather than in isolation, this provides a much more powerful approach to neuroimaging studies, yielding predictive results that are four times greater than when brain regions are tested in isolation,” says lead author author Tamas Spisak. , head of the Prognostic Neuroimaging Laboratory at the Institute for Diagnostic and Interventional Radiology and Neuroradiology at the University Medicine Essen.
However, not all pattern recognition algorithms are created equal, and finding which algorithms work best for specific types of brain imaging data is an active area of research. Previous work by Marek, Tervo-Clemmens et al. They also tested whether pattern recognition can be used to predict features from brain images, but Spisak and colleagues found that the algorithm they used was suboptimal.
When the researchers applied a more powerful algorithm, the results became even larger, and reliable associations could be detected in much smaller samples. “When you do the power calculations about how many participants are needed to detect replicable effects, the number drops below 500 people,” says Spisak.
“This opens the field to studies of many characteristics and clinical conditions for which it is not possible to obtain thousands of patients, including rare brain disorders,” says co-author Ulrike Bingel at University Medicine Essen, who heads the University Center for Pain. Medicine. “Identification of markers, including those involving the central nervous system, is urgently needed, as they are critical for improving diagnostic and personalized treatment approaches. We need to move towards a personalized medicine approach based on neuroscience. The potential of multivariate BWAS to move us towards this goal should not be underestimated.’
The team explains that the weak correlations found in the previous analysis, particularly through brain imaging, were collected while people were simply resting in the scanner, rather than performing tasks. But fMRI can also record the brain activity associated with specific moment-by-moment thoughts and experiences.
Wager believes that linking brain patterns to these experiences may be the key to understanding and predicting differences between individuals. “One of the challenges associated with using brain imaging to predict traits is that many traits are not stable or reliable. If we use brain imaging to focus on studying mental states and experiences, such as pain, empathy and drug cravings, the results can be much bigger and more reliable,” says Wager. “The key is finding the right project to conquer the state.”
“For example, viewing drug images in people with substance use disorders can induce drug cravings, according to an earlier study that revealed a neuromarker for craving,” says Wager.
“Determining which approaches to understanding the brain and mind are most likely to succeed is important because it affects how stakeholders view and ultimately fund translational research in neuroimaging,” says Bingel. “Finding the limitations and working together to overcome them is key to developing new ways to diagnose and care for patients with brain and mental health disorders.”